Vicidial Cluster

[luiz.avelino]

Hello Everyone.

I am installing vicicial version 5 in cluster to test and homologate so i can upgrade my environment.
For me more important in my environment of the cluster is web server and data base server of course.
After some tests, i checked that the data base server work with master slave.
But if the data base master fail and the data base slave assume, the data base slave receive request of the environment.
Now if my data base master be back, it will no synchronizes the data base with data base slave, that to the world of the Call Center not correct or not is ideal.
How i do to work with two data base synchronizes ?

Thanks.

[Acidshock]

I think the master/slave setup is more for performance than redundancy in VICIDIAL
s situation. The idea that you can run nasty queries that lock up tables for a while on the slave while the master still runs the center.

I may be wrong since I haven't setup a master/slave instance yet. Its on my todo list though. I would love to hear input on this one too.

[geoff3dmg]

Yes the primary purpose of the slave is for reporting (since it's read only). Some report queries can be quite taxing. However if your master does fail, it is possible to swap over to the slave (although some manual intervention is required).

[luiz.avelino]

Actually I did not want make manual has intervened, i wanted a environment where i not needed intervened daily, because i know that, if my environment data base fail, i will intervened manual.
Someome already work com data base master/master, i think that would be most ideal for me.
In my environment web server i will use heartbeat to make fail over, i already worked with heartbeat, i think this tool very good to failover, i think that in this case this all right.
The big problem is in environment data base server.
I wanted remember in Brazil, i think that few enterprise use vicidial, i think.
Thanks.

[geoff3dmg]

MariaDB has multi master support. However MyISAM database support is experimental.

https://mariadb.com/kb/en/what-is-maria ... a-cluster/

[mflorell]

I would NOT recommend using a Master-Master setup, there is a much greater chance of data corruption, and it's not really necessary with a good slave-failover process.

[GaD]

I've setup master/master/master replication (circular replication, 3 servers), but I point the write queries to only one server at a time, the others are used as read only, at least on the Vici DB. If the read/write DB goes down you can VERY QUICKLY move to the next server just by modifying the DB server IP, when the failed server comes back up it will sync the missing data and just wait for either a read or a write request to come in.... Pretty simple to manage but beware that you need to have good connection between nodes to ensure the lastest data to be spread throught the servers quickly.